Replace the thread used in the test case teardown with a Ticker to allow
testing with the bare metal profile. This Ticker is used to prevent the
watchdog from resetting the device during the final greentea
communication.
* Change "is supported" check to be a macro, so it can be done at
compile-time.
* Eliminate weird shift on 7-bit CRCs.
* Add support for 32-bit CRCs and reversals to TMPM3HQ.
This test is failed since #11721 Bare metal greentea support
on many STM32 boards:
NUCLEO_F072RB, NUCLEO_F070RB
NUCLEO_F411RE, NUCLEO_F439ZI, NUCLEO_F446ZE, DISCO_F407VG
NUCLEO_F303RE
...
* Change "is supported" check to be a macro, so it can be done at
compile-time.
* Eliminate weird shift on 7-bit CRCs.
* Add support for 32-bit CRCs and reversals to TMPM3HQ.
-Added "rtos-api" in TESTS/configs/baremetal.json
-some test case used thread_sleep_for instead Thisthread::sleep_for
-Added the EOF character
-Disable completely the non supported test cases
Extend the deepsleep wait up to 220% of the Watchdog timeout.
One of the current Watchdog timing requirements is to fire BEFORE twice
the timeout value, but at least one target (K64F) is expected to fire
EXACTLY at a doubled timeout value in deepsleep mode. This patch updates
the test to cope with a new deepsleep timing requirement.
Add a watchdog-kicking thread running in the background when the test
suite is handling a failed assertion. A single watchdog kick did not
provide enough time for the greentea communication if the watchdog
reset happened later than expected.
Notes:
- CRC and TRNG are disabled on this board since the MCU on this board does not have the required hardware.
- QSPI and QSPIF are temporarily removed due to issues related to supporting the new S25FS512S flash chip we have on this board. These will be enabled as soon as we can get the issues resolved.
https://github.com/ARMmbed/mbed-os/issues/11722
The QSPI spec allows alt to be any size that is a multiple of the
number of data lines. For example, Micron's N25Q128A uses only a
single alt cycle for all read modes (1, 2, or 4 bits depending on
how many data lines are in use).
* Fix rtos include path in NRFCordioHCIDriver
* Flatten USB driver directory structure
* Add missing include for us_ticker
* Add more missing includes for us_ticker
* Fix mbed_hal_fpga_ci_test_shield/uart test
* Fix bare-metal build
* Fix Watchdog UNITTEST
* Fix Mbed OS 2 build for Public/Internal headers relocating
Watchdog::stop() is called in this test. If the target does not support
stopping the watchdog this case needs to be skipped.
Also improve handling ignored test cases.
Update the watchdog timing requirements:
1. The watchdog should trigger at, or after the timeout value.
2. The watchdog should trigger before twice the timeout value.
Update the watchdog timing requirements:
1. The watchdog should trigger at, or after the timeout value.
2. The watchdog should trigger before twice the timeout value.
Remove explicit calls to sleep.
Update the watchdog timing requirements:
1. The watchdog should trigger at, or after the timeout value.
2. The watchdog should trigger before twice the timeout value.
Update the watchdog timing requirements:
1. The watchdog should trigger at, or after the timeout value.
2. The watchdog should trigger before twice the timeout value.
With the DEEP_SLEEP_TEST_CHECK_WAIT_DELTA_US increased,
we now have TEST_ASSERT_UINT64_WITHIN(delta=1000, expected=1000, actual=1000)
so this assertion needed to be updated.
What we need is the deep sleep to be enabled after the programed interrupt
has fired and before a 2ms timeout expiration, which means >= 1000 and < 2000.
Changes:
- restore the original form of setup/teardown handlers,
- test_lock_unlock_test_check(): do not use common ticker layer (Timer, Timeout). Use only ticker HAL layer.
- Increase DEEP_SLEEP_TEST_CHECK_WAIT_DELTA_US delta.
Added a mechanism which will prevent an initialised watchdog from
resetting the device during final greentea communication.
This allows testing watchdog timeouts as short as 50 ms.
As the timer code became more generic, coping with initialization on
demand, and variable width and speed us_ticker_api implementations,
wait_us has gradually gotten slower and slower.
Some platforms have reportedly seen overhead of wait_us() increase from
10µs to 30µs. These changes should fully reverse that drop, and even
make it better than ever.
Add fast paths for platforms that provide compile-time information about
us_ticker. Speed and code size is improved further if:
* Timer has >= 2^32 microsecond range, or better still is 32-bit 1MHz.
* Platform implements us_ticker_read() as a macro
* Timer is initialised at boot, rather than first use
The latter initialisation option is the default for STM, as this has
always been the case.
In particular and as kindly suggested by Przemec S. :
1. Add setup/teardown handler’s for all cases. This disables sys-tick,
so there should be no unexpected lp ticker interrupt scheduling.
2. Modify setup/teardown handler’s: remove suspension of lp/us tickers,
so they can count as this is required by test_lock_unlock_test_check test
case.
3. Use TEST_ASSERT_TRUE(sleep_manager_can_deep_sleep_test_check()) after
setting interrupt to cope with STM specific handling (CMPOK interrupt with
deep-sleep locked). This performs wait only if needed and will not affect
other targets which do not need extra wait.
4. Move sleep_manager_lock_deep_sleep() after TEST_ASSERT_TRUE(sleep_manager_can_deep_sleep_test_check())
5. Use TEST_ASSERT_TRUE(sleep_manager_can_deep_sleep_test_check()) in
test_lock_unlock_test_check to let lower layers manage deep sleep.
Time drifting test cases use serial communication with the host and are unstable on CI.
Skip time-drifting test cases if SKIP_TIME_DRIFT_TESTS macro is defined.
The idea for the future is to use FPGA test shield for timing tests instead of host scripts.
Also remove `__ARM_FM` macro which in most cases was used to disable time drifting tests. In other cases replace `__ARM_FM` with `TARGET_ARM_FM` which is more suitable.
On some targets like STM family boards with LPTIM enabled an interrupt is triggered on counter rollover.
We need special handling for cases when next_match_timestamp < start_timestamp (interrupt is to be fired after rollover).
In such case after first wake-up we need to reset interrupt and go back to sleep waiting for the valid one.
On some targets like STM family boards with LPTIM enabled there is a required delay (~100 us) before we are able to reprogram LPTIM_COMPARE register back to back.
This is handled by the low level lp ticker wrapper which uses LPTIM_CMPOK interrupt. CMPOK fires when LPTIM_COMPARE register can be safely reprogrammed again.
This means that on these platforms we have additional interrupt (CMPOK) fired always ~100 us after programming lp ticker.
Since this interrupt wake-ups the board from the sleep we need to go to sleep after CMPOK is handled.
Background:
There is an errata in LPTIM specification that explains that CMP Flag
condition is not an exact match (COUNTER = MATCH) but rather a
comparison (COUNTER >= MATCH).
As a consequence the interrupt is firing early than expected when
programing a timestamp after the 0xFFFF wrap-around.
In order to
work-around this issue, we implement the below work-around.
In case timestamp is after the work-around, let's decide to program the
CMP value to 0xFFFF, which is the wrap-around value. There would anyway be
a wake-up at the time of wrap-around to let the OS update the system time.
When the wrap-around interrupt happen, OS will check the current time and
program again the timestamp to the proper value.
Kevin Bracey
Martin Kojtal
Hugues Kamba
Filip Jagodzinski
Rajkumar Kanagaraj
jeromecoutant
Chris Trowbridge
Evelyne Donnaes
Alexandre Bourdiol
Kyle Kearney
Matthew Macovsky
Ryan Morse
Maciej Bocianski
int_szyk
Laurent Meunier
George Psimenos
Przemyslaw Stekiel
Seppo Takalo
Chun-Chieh Li
Anna Bridge
Filip Jagodzinski
Lingkai Dong